Merge pull request #4 from obsidiansystems/split-git

Split git

src/libutil/git.cc

@@ -51,53 +51,74 @@ static void parse(ParseSink & sink, Source & source, const Path & path) {
     else {
     }
 }
-// Internal version, returns the perm.
-unsigned int dumpGitInternal(const Path & path, Sink & sink, PathFilter & filter)
+
+// TODO stream file into sink, rather than reading into vector
+GitMode dumpGitBlob(const Path & path, const struct stat st, Sink & sink)
+{
+    auto s = (format("blob %d\0%s"s) % std::to_string(st.st_size) % readFile(path)).str();
+
+    vector<uint8_t> v;
+    std::copy(s.begin(), s.end(), std::back_inserter(v));
+    sink(v.data(), v.size());
+    return st.st_mode & S_IXUSR
+        ? GitMode::Executable
+        : GitMode::Regular;
+}
+
+GitMode dumpGitTree(const GitTree & entries, Sink & sink)
 {
+    std::string s1 = "";
+    for (auto & i : entries) {
+        unsigned int mode;
+        switch (i.second.first) {
+        case GitMode::Directory: mode = 40000; break;
+        case GitMode::Executable: mode = 100755; break;
+        case GitMode::Regular: mode = 100644; break;
+        }
+        s1 += (format("%6d %s\0%s"s) % mode % i.first % i.second.second.hash).str();
+    }
+
+    std::string s2 = (format("tree %d\0%s"s) % s1.size() % s1).str();
+
+    vector<uint8_t> v;
+    std::copy(s2.begin(), s2.end(), std::back_inserter(v));
+    sink(v.data(), v.size());
+    return GitMode::Directory;
+}
+
+// Returns the perm in addition
+std::pair<GitMode, Hash> dumpGitHashInternal(
+    std::function<std::unique_ptr<AbstractHashSink>()> makeHashSink,
+    const Path & path, PathFilter & filter)
+{
+    auto hashSink = makeHashSink();
     struct stat st;
+    GitMode perm;
     if (lstat(path.c_str(), &st))
         throw SysError(format("getting attributes of path '%1%'") % path);
 
     if (S_ISREG(st.st_mode)) {
-        auto s = (format("blob %d\0%s"s) % std::to_string(st.st_size) % readFile(path)).str();
-
-        vector<unsigned char> v;
-        std::copy(s.begin(), s.end(), std::back_inserter(v));
-        sink(v.data(), v.size());
-        return st.st_mode & S_IXUSR
-            ? 100755
-            : 100644;
-    }
-
-    else if (S_ISDIR(st.st_mode)) {
-        std::string s1 = "";
-
-        std::map<string, string> entries;
+        perm = dumpGitBlob(path, st, *hashSink);
+    } else if (S_ISDIR(st.st_mode)) {
+        GitTree entries;
         for (auto & i : readDirectory(path))
-            entries[i.name] = i.name;
-
-        for (auto & i : entries)
-            if (filter(path + "/" + i.first)) {
-                HashSink hashSink(htSHA1);
-                unsigned int perm = dumpGitInternal(path + "/" + i.first, hashSink, filter);
-                auto hash = hashSink.finish().first;
-                s1 += (format("%6d %s\0%s"s) % perm % i.first % hash.hash).str();
+            if (filter(path + "/" + i.name)) {
+                entries[i.name] = dumpGitHashInternal(makeHashSink, path + "/" + i.name, filter);
             }
-
-        std::string s2 = (format("tree %d\0%s"s) % s1.size() % s1).str();
-
-        vector<unsigned char> v;
-        std::copy(s2.begin(), s2.end(), std::back_inserter(v));
-        sink(v.data(), v.size());
-        return 40000;
+        perm = dumpGitTree(entries, *hashSink);
+    } else {
+        throw Error(format("file '%1%' has an unsupported type") % path);
     }
 
-    else throw Error(format("file '%1%' has an unsupported type") % path);
+    auto hash = hashSink->finish().first;
+    return std::pair { perm, hash };
 }
 
-void dumpGit(const Path & path, Sink & sink, PathFilter & filter)
+Hash dumpGitHash(
+    std::function<std::unique_ptr<AbstractHashSink>()> makeHashSink,
+    const Path & path, PathFilter & filter)
 {
-    dumpGitInternal(path, sink, filter);
+    return dumpGitHashInternal(makeHashSink, path, filter).second;
 }
 
 }

src/libutil/git.hh

@@ -16,6 +16,20 @@ void restoreGit(const Path & path, Source & source);
 
 void parseGit(ParseSink & sink, Source & source);
 
-void dumpGit(const Path & path, Sink & sink, PathFilter & filter = defaultPathFilter);
+// Dumps a single file to a sink
+GitMode dumpGitBlob(const Path & path, const struct stat st, Sink & sink);
 
+typedef std::map<string, std::pair<GitMode, Hash>> GitTree;
+
+// Dumps a representation of a git tree to a sink
+GitMode dumpGitTree(const GitTree & entries, Sink & sink);
+
+// Recursively dumps path, hashing as we go
+Hash dumpGitHash(
+    std::function<std::unique_ptr<AbstractHashSink>()>,
+    const Path & path,
+    PathFilter & filter = defaultPathFilter);
+
+// N.B. There is no way to recursively dump to a sink, as that doesn't make
+// sense with the git hash/data model where the information is Merklized.
 }

src/nix/hash.cc

@@ -36,14 +36,17 @@ struct CmdHash : Command
 
     std::string description() override
     {
-        const char* d;
+        const char *d;
         switch (mode) {
         case FileIngestionMethod::Flat:
             d = "print cryptographic hash of a regular file";
+            break;
         case FileIngestionMethod::Recursive:
             d = "print cryptographic hash of the NAR serialisation of a path";
+            break;
         case FileIngestionMethod::Git:
             d = "print cryptographic hash of the Git serialisation of a path";
+            break;
         };
         return d;
     }
@@ -78,9 +81,7 @@ struct CmdHash : Command
                 break;
             }
             case FileIngestionMethod::Git:
-                auto hashSink = makeHashSink();
-                dumpGit(path, *hashSink);
-                h = hashSink->finish().first;
+                h = dumpGitHash(makeHashSink, path);
                 break;
             }